Research Of Community Detection Based On Node Importance And Dynamic Community Evolution Model In Complex Networks

Studies concerning complex networks have indicated that they have the following properties: small-world network, scale-free network, robustness and vulnerability, community structure, and other characteristics. The improvement of the reliability and survivability of the network through the study of its robustness and vulnerability has become an important topic in recent years. Exploring and evaluating the importance of a node is highly valuable in various, practical applications. Hence, seeking and protecting vital nodes in the whole network is important. Many complex networked systems are found to divide naturally into modules or communities, groups of vertices with relatively dense connections within groups but sparser connections between them. The ability to find and analyze such groups can provide invaluable help in understanding and visualizing the structure of networks. For the modeling of complex networks it should have the following characteristics: global features and local characteristics. The former has the power-law characteristics, isomerism and heterogeneity, dynamic growth and regression of nodes; the latter has such as local priority connection and community structure.A new multi-index evaluation algorithm based on principal component analysis has been proposed. The algorithm presented in the current study synthesizes the topological characteristics of the network, such as degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, mutual information centrality, and subgraph centrality to reflect the relative node importance. In the next, we synthesize six statistical indexes in complex network to construct a multi-dimensional evaluation matrix which is converted by Gaussian kernel function, through the relationship we get a new comprehensive and balanced multidimensional index data by using the contribution rate of the eigenvalues as weights coefficient. The results indicated that he validity of the algorithm and the model.A new algorithm has been proposed based on self-adapted Fuzzy c-means clustering to detecting communities. The algorithm acquires the best community structure in complex networks via constructing a new validity function to find an optimal number of clusters voluntarily. In the next, on the basis of the model in Section3, by defining a concept which called community element similarity, constructing a matrix of the community element importance and analyzing the dominate relationships between them; we proposed a new community detection algorithm. To verify the validity of the algorithms, some simulation experiments were conducted. The results indicated that the algorithms are rational, effective, complete and accurate. A new evolution model of complex network from the view of hierarchical community structure has been established by synthesizing the current research status of the modeling of the complex network and combining with the above characteristics and based on BA scale-free network model. In the next, a dynamic evolution network model was built by establishing a new evolution rule of cellular automata based on the above evolution model. Results revealed that the network exhibits a power-law behavior and has a clear, controlled community structure. These models can suitably describe the characteristics of complex network topologies.